基于方向场的指纹图像偏微分方程修补模型

提出一个用于指纹图像修复的新的偏微分方程(PDE)模型,该模型对指纹图像的缺损区域能进行有效的修补。通过分析比较现有的技术方案对指纹图像修补的不足:一般常见的图像修复模型由于缺乏方向场的几何信息,对于指纹图像,这些模型不能给出满意的修补结果;或者虽然引入了方向场的几何信息,但在具体修补时会出现将不同的脊线连到一起不同的错误结果。该模型采用方向场作为扩散方向,在扩散过程中灰度信息沿着表征脊线方向的局部固定方向传播到待修复区域中, 改进了一般PDE模型不能用于修复指纹图像的不足。数值实验结果表明在修复指纹图像时提出的模型优于一般的模型

Partial differential equation model of fingerprint image inpaiting based on orientation field

This paper presented a new Partial Differential Equation (PDE) model for fingerprint image restoration, which was an effective method for fingerprint image automatic inpainting. The existing solutions to image inpainting have some drawbacks: satisfactory inpainting results for fingerprint image cannot be provided by common image inpainting models, usually due to the lack of geometric information of the direction field; or because of the introduction of geometric information of the direction field, error results, such as different ridge lines connected to each other, will appear during inpainting process. The main principle of the presented model was to employ the orientation field to act as the constraint of the diffusion directions after comparing and analyzing the existing solutions, and the gray information could be propagated into the inpainting domain along a local fix orientation in the inpainting process, which characterized the orientation of the ridges. The presented model improved normal PDE models for fingerprint image inpainting. The numerical experimental results show that the proposed model is superior to common models at inpainting fingerprint images.